Growing Up in Australia: The Longitudinal Study of Australian Children: A Valuable New Data Source for Economists

There is a long history of economic research into issues affecting children and families with children in areas, such as equivalence scales (Engel 1895), poverty (Rowntree 1901), and fertility and population growth (for example, Malthus 1826). Adam Smith (1776, p. 213), discussing colonies in North America, provided an early statement of the importance of investments in early childhood: The children, during the tender years of infancy, are well fed and properly taken care of, and when they are grown up, the value of their labour greatly overpays their maintenance. When arrived at maturity, the high price of labour, and the low price of land, enable them to establish themselves in the same manner as their fathers did before them. In the 1960s and 1970s, the interest in economic issues related to children and families with children grew rapidly. There were expansions into areas, such as human capital development (Becker 1965), female labour supply (Killingsworth 1983) and internal family dynamics (Manser and Brown 1980). More recently, economists have been using longitudinal studies of child development to address a range of economic questions, particularly questions at the intersection of economics and developmental psychology (for example, Hill et al. 2005; Waldfogel 2006). There also has been work undertaken on the economics of investment in early childhood (Carneiro and Heckman 2003). With the release of the data from the second wave of Growing Up in Australia: The Longitudinal Study of Australian Children (LSAC), Australia now joins most other Organisation for Economic Co-operation and Development countries in having a nationally representative longitudinal study of child development.1 Until the advent of the LSAC, the analysis of these issues for Australian children and families was limited by a lack of suitable data. Prior to the LSAC, Australian longitudinal studies of children's development usually focused on a narrow range of outcomes, were very health-focused and did not have nationally representative samples or had relatively small sample sizes. Generally, limited information was collected on factors, such as the parental labour force status or the economic position of households in which children live.2 From the perspective of an economist, these studies often are only of limited use and can be used only for addressing very specific questions. The LSAC aims to shed light on the development of the current generation of Australian children and to investigate the contribution of the children's social, economic and cultural environments to their adjustment and well-being. More specifically, it seeks to improve the understanding of the complex interplay of factors that foster or impede healthy early childhood development, to identify the opportunities for early intervention and prevention in policy areas concerning children and families, and to inform the policy debate in general. The study enables an examination of the impact of children's early experiences on their later development and the mapping of the diverse pathways that children follow as they develop. It also provides the opportunity to identify the long-term consequences of policy for children and their family. The survey collects information on multiple facets of children's development, health and well-being, including physical health and social, cognitive and emotional development. The study seeks to understand the risks and protective processes underlying children's development, looking at the interaction between children's psychological and physical attributes (such as their temperament and health) and the contexts in which children are raised, particularly their family, child care, school, neighbourhood and community experiences. The study also examines the dynamics within these settings; for example, the parenting practices, the quality of co-parental relationships to which the children are exposed and the quality of care received in differing types of non-parental care. The LSAC was initiated and is funded by the Department of Families, Housing, Community Services and Indigenous Affairs (FaHCSIA), Australian Government, and the study is managed in partnership with the Australian Institute of Family Studies (AIFS) and the Australian Bureau of Statistics (ABS). A consortium of leading researchers and experts from universities and research agencies provides advice on design and methodology issues. Wave 1 of the field work was conducted by I-View and the field work from Wave 2 onward is being undertaken by the ABS. The funding for four biennial waves was initially provided, with the first wave conducted in 2004. This funding subsequently has been extended to provide for a further four biennial survey waves. This review seeks to provide an overview of the LSAC and to highlight features of the data that are likely to be of particular interest to economists. The LSAC is following two cohorts of children for 14 years (possibly longer), starting from when the children were aged 0–1 year and 4–5 years. The 0–1 year old cohort is often described as the B (baby) cohort and the 4–5 year old cohort as the K (kindergarten) cohort. Alternatively, they can be identified by the years of their birth: the 1999–2000 birth cohort and the 2003–2004 birth cohort. Face-to-face interviews are being conducted every 2 years, with the first wave of data collection occurring in 2004 (March to November). The Wave 2 field work commenced in April 2006 and was completed in early 2007 and the Wave 3 field work commenced in April 2008 and was completed in early 2009. The two-cohort design will enable information on children's development over the first 18 years of life to be collected over 14 years. The two cohorts also can be compared at overlapping ages to gauge the effect of growing up in differing social conditions and policy settings. The overlapping of cohorts has occurred for the first time with the completion of Wave 3 (at ages 4–5 and 6–7 years) and will continue to occur for each subsequent wave. During 2004, the study recruited a nationally representative sample of 5107 infants and 4983 children aged 4–5 years, selected from the Medicare Australia enrolment database. The sampling unit is ‘the child’. A two-stage clustered design was employed, first selecting postcodes then children, allowing an analysis of children within their community and reducing the overall cost of the study. The children in both cohorts were selected from the same postcodes, with about 40 children per postcode usually invited to participate in the study in the larger states and usually about 20 children per postcode in the smaller states and territories. Stratification was used to ensure proportional geographical representation for the states and territories and capital city statistical division. The method of postcode selection took into account the number of children in the postcode; hence, all the potential participants in the study Australia-wide had an approximately equal chance of selection (about one in 25). Further information on the sampling is provided by Soloff et al. (2006). The sample is broadly representative of all Australian children (citizens and permanent residents) in each of two selected age cohorts: children born between March 2003 and February 2004 (B cohort) and children born between March 1999 and February 2000 (K cohort). The children in some remote parts of Australia were excluded because of the extremely high data collection costs in these areas. Reports from multiple participants are sought in order to obtain information about the child's behaviour across differing contexts and to reduce the effects of respondent bias. Information is being collected from the child (using physical measurement, cognitive testing and interviewing, depending upon the age of the child), the parents who live with the child (biological, adoptive or step-parents), home-based and centre-based carers for preschool children who are regularly in non-parental care and teachers (for school-aged children). From Wave 2, information also is being sought from parents who live apart from their child but who still have contact with the child (in the LSAC terminology, the ‘parent living elsewhere’). Table 1 provides a summary of the sources of information of the LSAC. Face-to-face interviews are conducted with the primary carer of the child (Parent 1). At Wave 1, this was the child's biological mother in 97 per cent of families. In addition, Parent 1 is asked to complete a questionnaire and a time use diary about the child's activities. The other resident parent (biological, adoptive or step-parent) is also asked to complete a leave-behind questionnaire. Consent is sought to send a self-completed questionnaire to the home-based carers, centre-based carers and teachers. Almost all the parents have provided consent. In addition, the interviewer has recorded some observations about the neighbourhood, family and child. In Wave 1, for the B cohort, physical measurements of the child were taken (weight and head circumference) and, for the K cohort, physical measurement and cognitive testing were carried out. In Waves 2 and 3, the height, weight and waist measurements were taken for both cohorts and age-appropriate cognitive testing continued. In Wave 3, the B cohort cognitive testing repeated the Wave 1 K cohort assessments. From the age of 6–7 years onwards, the child is interviewed. Although the focus of the LSAC is the study child, structured with the child as the unit of observation, the LSAC also can be used to conduct an analysis of the parents of the study child. As outlined above, the LSAC collects data on a wide range of topics. Information is collected on family demographics, including education, ethnicity and religion. Detailed information about the study child's physical and mental health, learning and cognition, and social and emotional outcomes is collected. How children develop is affected by their experiences in child care and school and the quality of the learning environment, so information is collected about these aspects of the study child's life. Quite detailed labour market information is collected about the parents and about the income and living standards of the family in which the child lives. A novel feature of the LSAC is the provision of time use diaries to collect data on a child's activities throughout two 24 hour periods, divided into 15 minute blocks. A summary of the topics covered by the LSAC and their scope is provided in Table 2. The final Wave 1 sample represented 53 per cent of all families who were sent a letter by Medicare Australia. Refusals were the largest source of sample loss (31 per cent for the B cohort and 35 per cent for the K cohort), followed by ‘non-contact’. Non-contact occurred when the address details were out-of-date or if only a post office box address was available. The rates of non-contact were 10 per cent for the B cohort and 14 per cent for the K cohort. The response rates achieved if non-contactable families are excluded is 64 per cent for the B cohort and 57 per cent for the K cohort.3 The Wave 2 data collection took place in 2006 (March to November), with 9070 families completing an interview (4606 in the B cohort and 4464 in the K cohort), yielding a response rate of 90 per cent of the originally recruited sample. Similar numbers of families from each cohort participated. The Wave 3 data collection took place in 2008. The response rate for Wave 3 was 86 per cent of the originally recruited samples, yielding a total of 8711 participants (4381 in the B cohort and 4330 in the K cohort). By the end of the Wave 3 data collection period, 460 families had withdrawn permanently from the study. A further 256 families elected not to participate in the third wave but remained members of the study. Ninety-three families were away for the entire fieldwork period and six children died since the start of the study. The remainder of the families could not be contacted. Thus, the largest cause of non-response was an inability to locate families. These families are still considered to be study members.4 This retention rate compares favourably with those achieved by comparable international studies. For example, the Millennium Cohort Study, a longitudinal study of approximately 19 000 families and babies in the United Kingdom, had a sample retention rate of 79 per cent between waves 1 and 2 over a similar time period to the LSAC (‘Sweep 1 and 2’ in the terminology used by the Millennium Cohort Study) (Hansen and Joshi 2007). The LSAC sample is broadly representative of the population, although there are some differences from the Census data of the ABS in relation to several family characteristics at Wave 1. Thus, the children with mothers or fathers who completed Year 12 are over-represented in the sample for both cohorts (differences of 8–10 percentage points). The infants with no siblings are under-represented (by 3 percentage points), while the 4–5 year olds in couple families are over-represented and those in sole-parent families are under-represented (each by 4 percentage points). The mothers who speak a language other than English at home are under-represented in both cohorts (by 2 percentage points). The AIFS (2008) provides a detailed comparison of the LSAC data with the census data. Weights are provided with the data to reduce the impact of biases in the sample selection process and survey non-responses. The clustered, stratified nature of the sampling needs to be taken into account when using the data. These issues are discussed in detail in Soloff et al. (2006). In Wave 2 of the LSAC, there was some variation in the participation rates of selected subgroups. Of single-parent families, 81 per cent took part, as did 80 per cent of Aboriginal and Torres Strait Islander families, 85 per cent of families in which the mother spoke English as a second language and 87 per cent of families in which the mother had not completed Year 12. The higher non-participation rate of these subgroups was primarily related to difficulties in making contact with the families, rather than their refusal to participate. Families can respond to all or only part of the study (with the exception of the face-to-face interview with Parent 1, which all families must complete in order to participate in the wave). Table 3 shows the response rates in Waves 1 and 2 for the differing data collection components. As well as the main data collection waves, between-wave mailed, self-completed surveys are conducted. The aim of these surveys is to gain an update of the child's progress in specific areas and the family's experiences, as well as to maintain the engagement of the study's families. To date, Waves 1.5 and 2.5 have been conducted. Wave 1.5 collected some information on the children's health and development, as well as their family, educational and care experiences. For the B cohort, information was collected on families’ service utilisation, parental employment history, maternity and other leave arrangements. Wave 2.5 collected information on work and child care, labour force status, child support and children's media and technology use. Wave 3.5 will occur in mid-2009. The between-wave mailed surveys have a lower response rate than do the main waves. Wave 1.5 resulted in 7167 useable responses of those who responded to Wave 1 (a response rate of 72 per cent). Wave 2.5 resulted in 6498 useable responses. The response rate of the families who had been interviewed in Wave 2 was 72 per cent, although forms also were received from 20 per cent of the families who had not participated in Wave 2 for whom there was contact information. Data from several administrative databases has been linked to the LSAC. These are the Medicare Australia database to obtain information about the child's medical service use and immunisation history and the National Childcare Accreditation Council data to obtain information on the child care centres attended by the children. Data from the Census of Population and Housing by the ABS on the socioeconomic and demographic profiles of the neighbourhoods and communities in which children live also has been linked to the LSAC data. As the study proceeds, it is hoped that further data sources will be linked to the LSAC; for example, the new national literacy and numeracy assessments and geospatial data sources. The LSAC dataset is publicly available, subject to an application process and the granting of a deed of licence. Data user training workshops are held regularly to assist users to gain familiarity with the complex dataset. The data from Waves 1, 1.5, 2 and 2.5 have been released and it is anticipated that the data from Wave 3 will be released towards the end of 2009. Information about accessing the data and user training is provided by the AIFS (2009). One of the strengths of the LSAC is the wide variety of areas of the children's lives about which the data is collected, as well as that of their family. This allows the inter-relationships between many different aspects of child development and family life to be analysed. Examples of issues of interest to economists into which the LSAC can provide insight include: The intergenerational linkages between parents and their children (Duncan and Brooks-Gunn 1997; Ermisch and Francesconi 2001). The particular value of the LSAC for addressing this question is that it provides information on the children's family, educational and child care experiences and economic circumstances from when the child is very young. The extent to which living in a disadvantaged neighbourhood impacts upon the outcomes for children over and above those of the family. This is a question on which there has been a great deal of international research (for example, Kling, Liebman and Katz 2007; Leventhal and Brooks-Gunn 2000), but little Australian research. The occurrence of changes in family structure (fertility, separation and divorce and repartnering) and the impacts of these changes on the short- and longer-term economic well-being of children (and other family members), as well as their developmental trajectories. Maternal, paternal and family labour supply decisions and the impacts of different patterns of labour force participation on children's developmental outcomes, as well as family relationships and functioning more broadly. This has been an active area of research internationally, but is yet to be analysed in any detail using Australian data. The economics of investment in early childhood. Investment in children's human capital and how this impacts upon the development of cognitive and non-cognitive skills. Child support and its impact on children's well-being and the economic position of their family. The impact of adverse events (for example, drought, unemployment, relationship breakdown, ill health or the death of a family member and housing stress) on the well-being of parents and children and the degree to which the impacts vary between children and families of different backgrounds. Although the LSAC is still young, it already has been used to analyse economic issues. Some examples are included to illustrate the ways in which the data have been and have used the LSAC to the cognitive between and has analysed the relationships between child outcomes and family socioeconomic The impact of neighbourhood on children's outcomes has been analysed by and and (2009). The data also have been used to analyse a range of labour market issues. The patterns of to work following and the impact of maternity leave have been the focus of several and and 2007; et al. 2007; and and a have analysed the labour market participation of mothers and fathers with children and the to well-being et al. 2007; and and and By the end of researchers will have to waves of an two between-wave mailed survey data. The value of the survey will only as waves of data are collected. The LSAC data for economic which only have to be